[gridengine users] Problems with quotas

[Sms Backup]

Dear all,

We have in our configuration multiple servers assigned to multiple queues.
To limit slots number per system, I tried to create qouta:

{
   name         slots
   description  Limit slots usage per node to number of cores
   enabled      FALSE
   limit        queues !interactive.q hosts {@allhosts} to slots=40
}

Expected result was to limit slots usage per host to 40 for non-interactive
queus, but allow interactive jobs even if all slots are taken.

It is doing it's job, but enabling this quota causes hundreds of thousands
messages from SGE master to be passed to systemd-journalctl, causing
systemd-journalctl eat 100% CPU, and make SGE master almost not usable,
example:

# systemctl status systemd-journald.service
● systemd-journald.service - Journal Service
   Loaded: loaded (/usr/lib/systemd/system/systemd-journald.service;
static; vendor preset: disabled)
   Active: active (running) since śro 2018-01-17 08:00:39 CET; 2 months 0
days ago
     Docs: man:systemd-journald.service(8)
           man:journald.conf(5)
 Main PID: 761 (systemd-journal)
   Status: "Processing requests..."
   CGroup: /system.slice/systemd-journald.service
           └─761 /usr/lib/systemd/systemd-journald

mar 19 10:26:20 qmaster systemd-journal[761]: Suppressed 783443 messages
from /system.slice/sgemas...ice
mar 19 10:26:50 qmaster systemd-journal[761]: Suppressed 767178 messages
from /system.slice/sgemas...ice
mar 19 10:27:20 qmaster systemd-journal[761]: Suppressed 823345 messages
from /system.slice/sgemas...ice
mar 19 10:27:50 qmaster systemd-journal[761]: Suppressed 831193 messages
from /system.slice/sgemas...ice
mar 19 10:28:20 qmaster systemd-journal[761]: Suppressed 684755 messages
from /system.slice/sgemas...ice
mar 19 10:28:50 qmaster systemd-journal[761]: Suppressed 669169 messages
from /system.slice/sgemas...ice
mar 19 10:29:20 qmaster systemd-journal[761]: Suppressed 659403 messages
from /system.slice/sgemas...ice

Did anyone had similar issue ? Any other ideas how to limit slots usage,
and not causing this kind of issue ? Or maybe it is possible to force SGE
not to send so many messages to systemd-journald ?

Regards,
Jakub

_______________________________________________
users mailing list
users@gridengine.org
https://gridengine.org/mailman/listinfo/users